The behavior of atmospheric models is sensitive to the algorithms that are used to represent the equations of motion. Typically, comprehensive models are conceived in terms of the resolved fluid dynamics (i.e. the dynamical core) and subgrid, unresolved physics represented by parameterizations. Deterministic weather predictions are often validated with feature-by-feature comparison. Probabilistic weather forecasts and climate projects are evaluated with statistical methods. We seek to develop model evaluation strategies that identify like “objects” – coherent systems with an associated set of measurable parameters. This makes it possible to evaluate processes in models without needing to reproduce the time and location of, for example, a particular observed cloud system. Process- and object-based evaluation preserves information in the observations by avoiding the need for extensive spatial and temporal averaging. As a concrete example, we focus on analyzing how the choice of dynamical core impacts the representation of precipitation in the Pacific Northwest of the United States, Western Canada, and Alaska; this brings attention to the interaction of the resolved and the parameterized components of the model. Two dynamical cores are considered within the Community Atmosphere Model. These are the Spectral (Eulerian), which relies on global basis functions and the Finite Volume (FV), which uses only local information. We introduce the concept of "meteorological realism" that is, do local representations of large-scale phenomena, for example, fronts and orographic precipitation, look like the observations? A follow on question is, does the representation of these phenomena improve with resolution?

Our approach to quantify meteorological realism starts with identification and isolation of key features of orographic precipitation that are represented differently by Spectral and FV models, using objective pattern recognition methods. Then we aim to quantitatively compare these features with observations (i.e. GPCC gauge based data) for validation of models. For this purpose we use methods of geospatial statistics. Specifically, we employ variography, which is a geostatistical method which is used to measure the spatial continuity of a regionalized variable, and principle component analysis which is an efficient method to extract trends in a dataset. We pose that these methods intrinsically link local, weather-scale phenomena to important climatological features and provide a quantitative bridge between weather and climate.